Improvement of the local prediction of chaotic time series

Abstract

In this paper we explore the effect of pseudofalse neighbor points, which are true neighbor points in the reconstructed attractor, but which are considered not suitable to be used when local methods are adopted to predict the chaotic time series. In our approach, the εp neighbor points are used to reduce the influence of the pseudofalse neighbor points, thereby improving the performance of the local prediction of the chaotic time series.